| *** annegentle has joined #openstack-swift | 00:06 | |
| *** dmorita has joined #openstack-swift | 00:07 | |
| *** edmondk has joined #openstack-swift | 00:21 | |
| edmondk | Quick question: Does anyone here know what the /opt/stack/data/swift/drives/images/swift.img file is for and does it ever clean itself up after removing swift elements? | 00:23 |
|---|---|---|
| clayg | probably never shrinks - its the file backed loop back device with all the devstack swift devices | 00:23 |
| notmyname | edmondk: is that something from devstack? (it's not an artifact created by code in the swift repo) | 00:23 |
| edmondk | Right now after uploading images to glance and creating swift elements and deleting my hard drive is still full even though no images exist | 00:24 |
| clayg | there's probably some devstack way to say resetswift | 00:24 |
| edmondk | notmmyname: yeah using devstack | 00:24 |
| *** bkopilov has quit IRC | 00:24 | |
| edmondk | does it hurt to just delete the file? | 00:24 |
| InAnimaTe | so...python-swiftclint....does it support 307's? | 00:24 |
| clayg | edmondk: it would probably only break your dev swift setup | 00:24 |
| InAnimaTe | (basically just had this discussion in #haproxy: http://tty0.in/view/c656172a) | 00:24 |
| notmyname | clayg: oh, is that a loopback mount fro swift in devstack? | 00:24 |
| edmondk | it has essentially filled the entire volume so I can't upload anymore images | 00:24 |
| clayg | edmondk: I get it - that sucks - you're going to have to delete it and recreate it - there's probably a way to tell devstack to do that - I don't know what it is | 00:25 |
| clayg | chmouel: acoles_away: cschwede: you guys know how to spell resetswift on devstack for edmondk | 00:25 |
| notmyname | InAnimaTe: unlikely. swift doesn't generate any 3xx responses. .....hmmm... /me wonders if the requests lib does it automatically | 00:26 |
| chmouel | edmondk: ./unstack.sh | 00:26 |
| chmouel | ./stack.sh | 00:26 |
| InAnimaTe | ok, im asing since im interested in using haproxy merely to help the client make a decision on which node to talk to, but i dont want any actual PUT's or GET's (or head's technically) going through the proxy, i just want it to be a broker | 00:27 |
| clayg | InAnimaTe: there's a couple of conversations going on in that thread - it sounds like you're going to make haproxy generate the 3XX - and if maybe swiftclient will fall for it? | 00:27 |
| InAnimaTe | yep | 00:27 |
| InAnimaTe | thats the goal | 00:27 |
| edmondk | Yeah figured unstack and stack would fix it, just without having to do that would be ideal because I have a bunch of instances running that were pretty hard to get setup right | 00:27 |
| clayg | InAnimaTe: i'm into it - probably won't work out of the box - have fun hacking - keep us posted | 00:28 |
| clayg | chmouel: thanks | 00:29 |
| clayg | edmondk: go digging in the devstack code - there's some lib function that makes that file - you could probably cook something up | 00:29 |
| edmondk | k | 00:29 |
| clayg | edmondk: yeah my money is on delete the file and unmount it all that jazz by hand - then try and run create_swift_disk in devstack/lib/swift | 00:30 |
| *** gyee has quit IRC | 00:31 | |
| *** annegentle has quit IRC | 00:31 | |
| openstackgerrit | Yuan Zhou proposed openstack/swift: List valid EC types in error https://review.openstack.org/142311 | 00:32 |
| edmondk | yeah I am in that func in devstack/lib/swift now | 00:32 |
| clayg | yuanz: doh! "on" | 00:32 |
| clayg | edmondk: sorry we couldn't be more help - most of us only use devstack when we have to test something with keystone | 00:32 |
| edmondk | clayg It looks like if I just call that method it checks for the directory and if found does a sudo unmount and sudo rm on it | 00:32 |
| clayg | phthhtppthtphtph - EVEN BETTER! | 00:33 |
| edmondk | clayg: thanks for the help will give that a go, if it fails I will go with the nuclear option and unstack and stack | 00:33 |
| tsg_ | torgomatic: jenkins is happy (https://review.openstack.org/#/c/163533/) is there a way to trigger "cluster-swift-tox-func"? | 00:35 |
| *** annegentle has joined #openstack-swift | 00:38 | |
| *** vishy has quit IRC | 00:41 | |
| *** vishy has joined #openstack-swift | 00:41 | |
| *** zhill has quit IRC | 00:44 | |
| *** tsg_ has quit IRC | 00:47 | |
| *** annegentle has quit IRC | 00:48 | |
| *** annegentle has joined #openstack-swift | 00:49 | |
| *** chlong has quit IRC | 00:55 | |
| *** chlong has joined #openstack-swift | 00:55 | |
| *** annegentle has quit IRC | 01:08 | |
| edmondk | clayg: I unmounted the file removed it, created a new swift.img and restarted swift but am unable to create containers now. I get Could not autocreate account '/AUTH_67bec053e00d4a24b205b083ccae7051' on s-proxy | 01:09 |
| edmondk | clayg: I am just going to go ahead and unstack and stack | 01:10 |
| clayg | yeah the swift.img needs to be loopbacked, partitioned and have filesystems on them, and mounted into the right places - it's a chore to do by hand and devstack knows how to do it somewhere | 01:11 |
| clayg | i thought getting that function to run might have put things back in the right place | 01:11 |
| clayg | next time I have my devstack box up I'll try and remember to take stab at it - put something in the interwebz for the next guy | 01:11 |
| peluse | torgomatic, clayg : tsg's pyeclib patch is ready for 2nd review (so EC patches can pass Jenkins)... https://review.openstack.org/#/c/163533/ | 01:13 |
| peluse | yuanz, thanks again for jumping back on the EC stuff to help our beta push! | 01:13 |
| *** clds_ has quit IRC | 01:42 | |
| *** clds has joined #openstack-swift | 01:44 | |
| yuanz | peluse, no problem, it's my honor :) | 01:55 |
| *** haomaiwang has joined #openstack-swift | 02:22 | |
| *** david-lyle is now known as david-lyle_afk | 02:45 | |
| *** echevemaster has quit IRC | 02:53 | |
| *** bkopilov has joined #openstack-swift | 02:58 | |
| notmyname | dfg: redbo: I'd like for https://review.openstack.org/#/c/126923/ to land. please take a look if you're interested (fsync() on dirs) | 03:15 |
| *** bkopilov has quit IRC | 03:19 | |
| clayg | hugokuo: charz: the qa cluster is real hardware right - can we add a ssbench run to the checks? | 03:26 |
| notmyname | clayg: yes, and probably. might need to rune them from the jenkins node | 03:26 |
| notmyname | clayg: what do you want to see? a raw number or comparison over time? | 03:26 |
| clayg | oh does the swiftqa fabric scripts not control what acctually gets run | 03:27 |
| notmyname | fabric like .fab files? is that how charz set it up? | 03:27 |
| notmyname | I think it's the jenkins gerrit plugin | 03:28 |
| clayg | i saw something in console.log for github.com/swiftstack/swiftqa | 03:28 |
| notmyname | but I don't really know | 03:28 |
| clayg | fab start_services | 03:29 |
| charz | clayg: yeah, I use fabric to do that. | 03:29 |
| clayg | so but who is acctually deciding to run tox to kick off the func tests? is that something you have to co-ordinate with infra? | 03:30 |
| clayg | or do they just like - here's a new change - do your thing | 03:30 |
| notmyname | clayg: the jenkins gerrit plugin is sniffing all the events from gerrit. and it acts on what it wants | 03:31 |
| charz | clayg: jenkins gerrit plugin did. | 03:31 |
| clayg | ok, so we have our own jenkins server, and we can make any jobs we want - then they get triggered by the gerrit hook plugin - i love it | 03:32 |
| clayg | so yeah - let's add an ssbench run job :D | 03:32 |
| notmyname | clayg: `ssh -p 29418 USERNAME@review.openstack.org gerrit stream-events` | 03:32 |
| clayg | all that jazz is all up in your anisble's anyway - doesn't have to be anything fancy | 03:32 |
| charz | clayg: jenkins server 192.168.201.155:8080 | 03:33 |
| clayg | hell yeah | 03:33 |
| notmyname | clayg: incoming.... | 03:35 |
| charz | clayg: notmyname How about add 1k/1m/10m PUT/GET benchmark? or we need more? | 03:37 |
| notmyname | charz: just for fun, I'd say run it with deletes turned off. at least for a while. that will gradually fill up the cluster | 03:38 |
| clayg | charz: that'd be fine - but it looks like it'd be easier to just add it to the existing job than to create a new job? | 03:38 |
| clayg | notmyname: there's a clean script that wipes disk | 03:38 |
| notmyname | new job, please | 03:38 |
| notmyname | clayg: I don't /think/ that's being run | 03:38 |
| notmyname | charz: ? | 03:38 |
| clayg | notmyname: I *just* said it'd be *easier* to add it the existing job - and you *immediately* ask for a new job :P | 03:39 |
| notmyname | lol | 03:39 |
| notmyname | in my experience, it's trivial to add a new job. at the end of the day, I mention it in passing to charz, and it's there by the time I wake up ;-) | 03:40 |
| clayg | +1 charz is amazing | 03:41 |
| charz | notmyname: clayg I'll create a new job for it. | 03:44 |
| notmyname | thanks. makes it easier to track and manage | 03:44 |
| *** ppai has joined #openstack-swift | 03:44 | |
| mattoliverau | notmyname, clayg: So from what I read from your conversation, charz is like a cross between a magic genie and tooth fairy? you wish for something, go to sleep and when you wake up its done :P | 04:01 |
| clayg | peluse: acoles_away: ok, i'm going to push up a rebase | 04:07 |
| charz | mattoliverau: like a ninja. :P | 04:07 |
| clayg | i tried to squash all the policy cleanup into multi-fragment but it was super obvious writing the commit the patch was just trying to do too many things at once | 04:08 |
| clayg | so I rebased policy-everywhere in front of the suffix changes (I think it made the suffix change simpler) | 04:08 |
| clayg | anyway - if it doesn't work out - that's why we use version control right! | 04:08 |
| clayg | charz: *thats* it! a *ninja* - i was trying to find the right word... | 04:09 |
| mattoliverau | lol | 04:10 |
| charz | LoL | 04:10 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Multiple fragment Archive Index Support https://review.openstack.org/159637 | 04:12 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Prefer policy to policy_index https://review.openstack.org/163690 | 04:12 |
| mattoliverau | sigh.. just as I was in the middle of reviewing these :P | 04:12 |
| clayg | mattoliverau: https://review.openstack.org/#/c/163690/1 will be easy | 04:14 |
| mattoliverau | clayg: famous last words :P | 04:16 |
| *** devlaps has quit IRC | 04:25 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Multiple fragment Archive Index Support https://review.openstack.org/159637 | 04:29 |
| *** erlon has quit IRC | 04:41 | |
| *** bkopilov has joined #openstack-swift | 04:56 | |
| *** david-ly_ has joined #openstack-swift | 05:04 | |
| *** yuanzz has joined #openstack-swift | 05:04 | |
| *** jd__` has joined #openstack-swift | 05:05 | |
| *** thurloat_ has joined #openstack-swift | 05:09 | |
| *** yuanz has quit IRC | 05:09 | |
| *** david-lyle_afk has quit IRC | 05:09 | |
| *** cebruns has quit IRC | 05:09 | |
| *** jd__ has quit IRC | 05:09 | |
| *** mariusv has quit IRC | 05:09 | |
| *** thurloat has quit IRC | 05:09 | |
| *** ho__ has quit IRC | 05:09 | |
| *** thurloat_ is now known as thurloat | 05:09 | |
| *** jd__` is now known as jd__ | 05:09 | |
| *** BAKfr has quit IRC | 05:09 | |
| *** redbo has quit IRC | 05:09 | |
| *** ho__ has joined #openstack-swift | 05:09 | |
| *** mariusv has joined #openstack-swift | 05:09 | |
| *** mariusv has quit IRC | 05:09 | |
| *** mariusv has joined #openstack-swift | 05:09 | |
| *** cebruns has joined #openstack-swift | 05:09 | |
| *** redbo has joined #openstack-swift | 05:09 | |
| *** ChanServ sets mode: +v redbo | 05:09 | |
| *** BAKfr has joined #openstack-swift | 05:13 | |
| *** reed has quit IRC | 05:30 | |
| *** SkyRocknRoll has joined #openstack-swift | 06:04 | |
| *** mmcardle has joined #openstack-swift | 06:20 | |
| *** silor has joined #openstack-swift | 06:21 | |
| *** mmcardle has quit IRC | 06:26 | |
| *** nshaikh has joined #openstack-swift | 06:29 | |
| *** ho__ has quit IRC | 06:46 | |
| *** ho has quit IRC | 07:14 | |
| *** silor has quit IRC | 07:16 | |
| *** SkyRocknRoll has quit IRC | 07:31 | |
| *** sluo_wfh has joined #openstack-swift | 07:42 | |
| *** sluo_wfh has quit IRC | 07:53 | |
| *** jistr has joined #openstack-swift | 08:03 | |
| *** sluo_wfh has joined #openstack-swift | 08:18 | |
| *** geaaru has joined #openstack-swift | 08:24 | |
| *** rledisez has joined #openstack-swift | 08:32 | |
| *** jordanP has joined #openstack-swift | 08:33 | |
| *** acoles_away is now known as acoles | 08:56 | |
| *** nellysmitt has joined #openstack-swift | 09:02 | |
| *** mmcardle has joined #openstack-swift | 09:30 | |
| *** SkyRocknRoll has joined #openstack-swift | 09:51 | |
| *** jordanP has quit IRC | 10:00 | |
| *** BAKfr has quit IRC | 10:35 | |
| *** BAKfr has joined #openstack-swift | 10:35 | |
| *** jistr has quit IRC | 10:36 | |
| m_han | hey guys ! i'm having an issue, my i'cant start my proxy-server : "swift-init main start gives the following exception http://paste.openstack.org/show/191794/ | 10:38 |
| m_han | this is the proxy-server.conf http://paste.openstack.org/show/191795/ any suggestions? thanx & best regards | 10:40 |
| *** BAKfr has quit IRC | 10:44 | |
| *** BAKfr has joined #openstack-swift | 10:44 | |
| *** haomaiwang has quit IRC | 10:48 | |
| omame | m_han: port already in use? | 10:51 |
| *** jistr has joined #openstack-swift | 10:53 | |
| m_han | omame: yes, it is listenning on tcp, but i have neither tomcat or apache running on my machine | 10:57 |
| omame | sudo netstat -ntlp | grep :8080 should give you a hint | 10:58 |
| omame | maybe it's proxy-server itself :) | 10:58 |
| m_han | tcp 0 0 127.0.0.1:8080 0.0.0.0:* LISTEN 1804/python | 10:58 |
| omame | cause you're starting, not restarting | 10:58 |
| omame | try swift-init proxy-server restart | 10:58 |
| *** erlon has joined #openstack-swift | 10:58 | |
| m_han | i get the same error http://paste.openstack.org/show/191797/ | 11:01 |
| omame | who's process 1804? | 11:02 |
| m_han | it was /usr/bin/python /usr/local/bin/swift-proxy-server /etc/swift/proxy-server.conf | 11:03 |
| m_han | i killed it and it works fine now xD | 11:03 |
| omame | ;) | 11:03 |
| m_han | thank u for ur support ;) | 11:04 |
| omame | no problem | 11:05 |
| openstackgerrit | Daisuke Morita proposed openstack/swift: object-replicator warns when object dirs mismatch with policy settings https://review.openstack.org/163389 | 11:08 |
| *** jrichli has joined #openstack-swift | 11:17 | |
| *** silor has joined #openstack-swift | 11:18 | |
| *** aix has joined #openstack-swift | 11:19 | |
| *** km has quit IRC | 11:20 | |
| *** nellysmi_ has joined #openstack-swift | 11:24 | |
| *** nellysmitt has quit IRC | 11:26 | |
| *** mahatic has joined #openstack-swift | 11:39 | |
| openstackgerrit | Denis Cavalcante proposed openstack/swift: Add storage policy support for sorting method https://review.openstack.org/160877 | 11:39 |
| *** nshaikh has quit IRC | 11:43 | |
| *** dmorita has quit IRC | 11:48 | |
| *** mahatic has quit IRC | 11:51 | |
| *** aix has quit IRC | 11:59 | |
| *** tsg_ has joined #openstack-swift | 11:59 | |
| *** mahatic has joined #openstack-swift | 12:06 | |
| *** mahatic has quit IRC | 12:13 | |
| *** mahatic has joined #openstack-swift | 12:15 | |
| *** silor has quit IRC | 12:25 | |
| *** SkyRocknRoll has quit IRC | 12:25 | |
| *** panbalag has joined #openstack-swift | 12:34 | |
| *** SkyRocknRoll has joined #openstack-swift | 12:39 | |
| *** mmcardle has quit IRC | 12:41 | |
| *** ahale is now known as _ahale | 12:49 | |
| *** NM has joined #openstack-swift | 12:50 | |
| *** jkugel has joined #openstack-swift | 12:57 | |
| *** aix has joined #openstack-swift | 12:58 | |
| *** bkopilov has quit IRC | 13:01 | |
| acoles | tsg_: hi you here? | 13:02 |
| tsg_ | acoles: yes, how is it going | 13:02 |
| acoles | tsg_: good, thanks for trying those rechecks - saw your note about rebasing, i will look at that again later today. thx | 13:03 |
| *** panbalag has quit IRC | 13:05 | |
| straycat | Why do openstack things tend to have bind_ip rather than bind_interface? | 13:06 |
| straycat | e.g. proxy-server.conf | 13:06 |
| tsg_ | acoles: yes, this exercise has been a little troublesome :) given we only have ubuntu packages built for the libs so far, we have had to get a special version of pyeclib up on pypi. Jenkins was unhappy yesterday because of some venv related issues that caused the early failures. Those were fixed later in the afternnon PDT and pyeclib 1.0.3 version should work just fine now. | 13:06 |
| acoles | tsg_: ok thx | 13:06 |
| tsg_ | acoles: clayg was having some issues last night - once those are resolved and he clears https://review.openstack.org/#/c/163533, we should be good for a rebase | 13:08 |
| tsg_ | acoles: would like to hear from you if you see any issues after you update pyeclib (and liberasurecode by proxy) | 13:08 |
| tsg_ | acoles: would like to get any caveats noted in the pyeclib README | 13:09 |
| *** SkyRocknRoll has quit IRC | 13:09 | |
| *** SkyRocknRoll has joined #openstack-swift | 13:10 | |
| *** mmcardle has joined #openstack-swift | 13:13 | |
| *** bill_az has joined #openstack-swift | 13:20 | |
| *** mmcardle has quit IRC | 13:24 | |
| *** mmcardle has joined #openstack-swift | 13:26 | |
| *** annegentle has joined #openstack-swift | 13:37 | |
| acoles | tsg_: so do we need to update requirements.tx for pyeclib>=1.0.3 ? | 13:41 |
| tsg_ | acoles: that's correct | 13:41 |
| tsg_ | acoles: I need to get that change into global-requirements and then on swift master | 13:42 |
| acoles | tsg_: ok ic | 13:42 |
| tsg_ | acoles: waiting on some feedback today to make sure we don't need a 1.0.4 :) | 13:43 |
| tsg_ | I don't anticipate changes in the core library, only some documentation to clarify dependencies better, etc | 13:45 |
| *** jrichli has quit IRC | 13:53 | |
| acoles | tsg_: so i pulled patch 163533, pip installed pyeclib 1.0.3, looks like i got liberasurecode 1.0.4, but see exception in unit test http://paste.openstack.org/show/191836/ - any ideas? | 13:55 |
| patchbot | acoles: https://review.openstack.org/#/c/163533/ | 13:55 |
| *** jrichli has joined #openstack-swift | 13:55 | |
| openstackgerrit | Prashanth Pai proposed openstack/swift: Refactor server side copy as middleware https://review.openstack.org/156923 | 13:56 |
| tsg_ | acoles: does /var/log/syslog show anything | 13:58 |
| acoles | tsg_: ah! Mar 12 13:51:36 localhost liberasurecode[11835]: liberasurecode_backend_open: dynamic linking error libgf_complete.so.1: cannot open shared object file: No such file or directory | 13:59 |
| tsg_ | acoles: we are trying to get rid of this limitation (syslog only logging) but given liberasurecode is a C library, we couldn't easily unify the logs :) | 14:00 |
| tsg_ | that apart, do you have libgf_complete installed under /usr/local/lib? | 14:00 |
| tsg_ | acoles: perhaps run ldconfig? | 14:00 |
| * tsg_ adding these notes to a "troubleshooting" section in pyeclib docs | 14:01 | |
| *** Fin1te has joined #openstack-swift | 14:02 | |
| acoles | tsg_: i have libgf_complete.so.1.0.0 , ldconfig seems to have done the trick, just running complete suite of unit test | 14:02 |
| tsg_ | great | 14:03 |
| *** ppai has quit IRC | 14:03 | |
| tsg_ | acoles: I am thinking that may have been clayg's problem as well :) | 14:04 |
| acoles | tsg_: ok all unit tests pass now. thanks! | 14:05 |
| tsg_ | acoles: good to know! I will get this documented | 14:05 |
| acoles | tsg_: i think (can't be sure) i had symlink to libgf_complete.so.1.0.0 before running ldconfig. | 14:05 |
| tsg_ | acoles: given we use autoconf+libtool for those libs, you shouldn't have to, but I will double check | 14:10 |
| *** Fin1te has quit IRC | 14:14 | |
| *** Fin1te has joined #openstack-swift | 14:14 | |
| *** Fin1te has quit IRC | 14:15 | |
| acoles | tsg_: so looks like maybe patch 163690 needs to be rebased onto 163533 and then the chain above that will pick up this fix? | 14:16 |
| patchbot | acoles: https://review.openstack.org/#/c/163690/ | 14:16 |
| tsg_ | acoles: that's correct | 14:18 |
| acoles | tsg_: ok i will try that | 14:18 |
| acoles | peluse: you awake yet? | 14:19 |
| tsg_ | acoles: peluse may be on a plane to SF | 14:19 |
| acoles | tsg_: he's awake then :) | 14:19 |
| tsg_ | acoles: :) heh, yes | 14:20 |
| acoles | tsg_: iirc he has a private jet ;) | 14:21 |
| tsg_ | acoles: yeah ;) and he is nice enough to lend it to a company named "Intel" sometimes | 14:22 |
| tsg_ | heh | 14:22 |
| acoles | tsg_: lol. he's good to them | 14:23 |
| *** lcurtis has joined #openstack-swift | 14:43 | |
| *** SkyRocknRoll has quit IRC | 14:58 | |
| *** csmart has quit IRC | 15:03 | |
| *** csmart has joined #openstack-swift | 15:04 | |
| *** rdaly2 has joined #openstack-swift | 15:04 | |
| *** bkopilov has joined #openstack-swift | 15:07 | |
| *** lcurtis has quit IRC | 15:10 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Prefer policy to policy_index https://review.openstack.org/163690 | 15:23 |
| *** david-ly_ is now known as david-lyle | 15:23 | |
| acoles | clayg: ^^ fixed pep8 issues and rebased on patch 163533 to try to get tests to pass, hope thats ok! | 15:24 |
| patchbot | acoles: https://review.openstack.org/#/c/163533/ | 15:24 |
| *** nellysmi_ has quit IRC | 15:24 | |
| tsg_ | acoles, thanks for the note on tox. btw, there is also an eventlet update sitting in the queue https://review.openstack.org/#/c/145403/ | 15:25 |
| *** thumpba has joined #openstack-swift | 15:26 | |
| tsg_ | acoles, jenkins hasn't been happy with these given these changes are submitted against feature/ec and there is no such branch in the global requirements repo | 15:27 |
| tsg_ | acoles, will need to get these updates submitted to swift master | 15:27 |
| acoles | notmyname: i noticed that openstackgerrit is not reporting feature/crypto reviews in this channel - who should i talk to about fixing that? | 15:29 |
| *** thumpba_ has joined #openstack-swift | 15:32 | |
| *** reed has joined #openstack-swift | 15:34 | |
| *** thumpba has quit IRC | 15:35 | |
| tsg_ | acoles: I am going to work on making pyeclib/liberasurecode logging configurable so we can control where those messages go from swift | 15:39 |
| tsg_ | acoles: please do let me know any suggestions you may have | 15:39 |
| *** zul has quit IRC | 15:40 | |
| *** zul has joined #openstack-swift | 15:41 | |
| *** nermik has joined #openstack-swift | 15:57 | |
| nermik | Hi! newbie question - does swift support partial file retrieval? if I define byte range for media file.. | 15:59 |
| acoles | nermik: yes with a Range header http://developer.openstack.org/api-ref-objectstorage-v1.html#storage_object_services | 16:02 |
| nermik | and if media is chunked and I need data from number of chunks - should I make number of calls and then "glue" chunks together ? | 16:08 |
| nermik | there is still 5GB object limitation ? | 16:08 |
| openstackgerrit | Lorcan Browne proposed openstack/swift: Add swift-recon feature to track swift-drive-audit error count https://review.openstack.org/163889 | 16:09 |
| ctennis | yes and yes and http://docs.openstack.org/developer/swift/overview_large_objects.html | 16:09 |
| *** thumpba has joined #openstack-swift | 16:12 | |
| nermik | ctennis, acoles: thanks! | 16:12 |
| ctennis | you're welcome! | 16:13 |
| *** dencaval has joined #openstack-swift | 16:14 | |
| *** bkopilov has quit IRC | 16:15 | |
| *** thumpba_ has quit IRC | 16:16 | |
| *** tsg_ has quit IRC | 16:17 | |
| *** bkopilov has joined #openstack-swift | 16:17 | |
| *** bkopilov has quit IRC | 16:20 | |
| *** thumpba_ has joined #openstack-swift | 16:21 | |
| m_han | hi, how to delete an object from swift-cluster ? | 16:23 |
| *** thumpba has quit IRC | 16:23 | |
| ctennis | m_han: issue an HTTP DELETE to the object | 16:24 |
| m_han | ctennis: it says http://paste.openstack.org/show/191870/ | 16:27 |
| ctennis | I don't think you're authenticated, you have to also pass your auth token as part of the request | 16:27 |
| ctennis | also that doesn't look like a valid url | 16:28 |
| ctennis | the /endpoints isn't right | 16:28 |
| *** bsdkurt has joined #openstack-swift | 16:28 | |
| m_han | the url is fine http://paste.openstack.org/show/191872/ | 16:35 |
| m_han | may be i'm missing another option in the curl command | 16:35 |
| m_han | any ideas? | 16:36 |
| *** jistr has quit IRC | 16:45 | |
| ctennis | no, the endpoint url you are using is hitting a different API | 16:47 |
| ctennis | if you were doing an actual GET on the object, it would return the contents and not the endpoints | 16:47 |
| ctennis | the object url is /v1/account/container/object | 16:47 |
| *** kirubak has joined #openstack-swift | 16:54 | |
| *** edmondk has left #openstack-swift | 16:55 | |
| notmyname | good morning | 17:19 |
| *** Nadeem has joined #openstack-swift | 17:22 | |
| *** aix has quit IRC | 17:25 | |
| *** rledisez has quit IRC | 17:26 | |
| *** devlaps has joined #openstack-swift | 17:26 | |
| *** kirubak has quit IRC | 17:32 | |
| *** nermik has quit IRC | 17:36 | |
| jrichli | notmyname: morning. good trip? | 17:50 |
| notmyname | yup. it was great to see people | 17:51 |
| notmyname | I like the ops summits. the sysadmins get an opportunity to talk and say what they really think without a lot of the traditional loud voices in openstack drowning them out | 17:51 |
| jrichli | :) | 17:52 |
| notmyname | and I was told on 2 different occasions this week that what the ops people are asking for is generally stuff that swift has always done. so that's nice :-) | 17:53 |
| jrichli | awesome! | 17:53 |
| *** geaaru has quit IRC | 17:54 | |
| *** zhill has joined #openstack-swift | 17:59 | |
| *** annegentle has quit IRC | 18:00 | |
| *** annegentle has joined #openstack-swift | 18:12 | |
| *** annegentle has quit IRC | 18:15 | |
| *** annegentle has joined #openstack-swift | 18:15 | |
| *** bill_az has quit IRC | 18:21 | |
| *** bkopilov has joined #openstack-swift | 18:21 | |
| *** tsg_ has joined #openstack-swift | 18:23 | |
| *** fbo has quit IRC | 18:23 | |
| notmyname | acoles: jrichli: I'll look in to getting the feature/crypto branch known by the patch bot | 18:24 |
| jrichli | thanks! | 18:25 |
| acoles | notmyname: thx | 18:25 |
| * notmyname remembers something about a yaml file being updated | 18:25 | |
| *** nermik has joined #openstack-swift | 18:26 | |
| *** fbo has joined #openstack-swift | 18:26 | |
| notmyname | yup | 18:27 |
| *** fbo has quit IRC | 18:30 | |
| notmyname | acoles: jrichli: https://review.openstack.org/#/c/163927/ | 18:30 |
| clayg | tsg_: torgomatic: acoles: yuanzz: so `sudo ldconfig` fixed things for me - i already had 1.0.4 liberasure installed? | 18:31 |
| torgomatic | clayg: that's good | 18:31 |
| acoles | clayg: same for me | 18:32 |
| torgomatic | I imagine this whole thing will get easier once pyeclib gets version checking installed | 18:32 |
| tsg_ | pyeclib does have version checking in fact | 18:32 |
| tsg_ | so it would install the right version (update if necessary) | 18:32 |
| tsg_ | in clayg's case, it must have been just the missing ldconfig | 18:32 |
| torgomatic | well, there you go :) | 18:33 |
| tsg_ | torgomatic, clayg: liberasurecode is configured to send error messages to syslog by default | 18:33 |
| tsg_ | I am trying to change that to also log to stderr, do you have a suggestion there? | 18:34 |
| clayg | tsg_: so the user having to run ldconfig is expected? | 18:34 |
| torgomatic | tsg_: wait, so under my Python code is a Python library and then under that is a C library that talks to my syslog? | 18:34 |
| clayg | tsg_: torgomatic: did all of you just *know* that and I'm some sort of johnny? | 18:34 |
| *** fbo has joined #openstack-swift | 18:34 | |
| torgomatic | clayg: the ldconfig thing? it said it on my terminal after the upgrade, so I did it | 18:35 |
| clayg | heh - i don't read that shit ;) | 18:35 |
| notmyname | acoles: jrichli: and it's already approved. as soon as it lands, the bot will auto-restart and it will be ready | 18:35 |
| torgomatic | clayg: evidently ;) | 18:35 |
| *** nellysmitt has joined #openstack-swift | 18:35 | |
| jrichli | notmyname: great! now if only I will be ready to upload my latest in a day or so :) | 18:37 |
| tsg_ | torgomatic: on the syslog comment, yes .. liberasurecode (C) is talking to syslog (we didn't change the defaults, but can do so) | 18:37 |
| clayg | acoles: so per policy diskfiles - I'm down for all the same reasons I think ECObjectController was a good idea | 18:37 |
| tsg_ | torgomatic: what would be the ideal way? stderr? or configure both pyeclib/liberasurecode to swift logs? | 18:38 |
| clayg | acoles: I think i want to add an distutils entrypoint for storage policy classes - and have them register their objectcontroller and diskfile in their factory method | 18:38 |
| clayg | acoles: but before we need that we can just have the ECStoragePolicy sort of do it in tree | 18:39 |
| acoles | clayg: are we talking objectcontroller as in backend server or proxy controller class? | 18:40 |
| clayg | I dream of a merge to master that is almost entirely *additive* like - here's the controller, and the diskfile, and it's off in it's own little sandbox, and very little shared code has to change except for where we add the routing hooks to lookup the controller and diskfile based on the policy | 18:40 |
| clayg | acoles: yeah if you look in feature/ec's proxy.controllers.obj - there's already a ECObjectController class | 18:40 |
| acoles | clayg: yeah, so are you also thinking ECObjectController in swift.obj.server ? | 18:41 |
| acoles | am i confused? | 18:41 |
| *** zul has quit IRC | 18:41 | |
| clayg | acoles: oh... i wasn't thinking about that :\ | 18:41 |
| *** fbo has quit IRC | 18:41 | |
| clayg | The difference with the storage server controllers and the proxy controllers is that a proxy controller instance gets created for every request | 18:41 |
| acoles | ok i *was* confused :) | 18:41 |
| acoles | clayg: yeah, ic | 18:42 |
| clayg | acoles: so part of the change was that I was thinking a single object server could service requests for different diskfiles | 18:42 |
| acoles | clayg: yes | 18:43 |
| clayg | so there'd have to be a "router" step - probably in object.server.ObjectController.get_diskfile | 18:43 |
| *** zul has joined #openstack-swift | 18:43 | |
| acoles | clayg: yup | 18:43 |
| *** silor has joined #openstack-swift | 18:44 | |
| clayg | acoles: top priority for me atm is to get the module level functions doing hash/suffix tracking moved to something more OO | 18:44 |
| acoles | clayg: so, i feel i'm lifting a bit of a stone with this pathc 162717 | 18:45 |
| acoles | patch 162717 | 18:45 |
| patchbot | acoles: https://review.openstack.org/#/c/162717/ | 18:45 |
| clayg | fucking colloquialisms | 18:46 |
| acoles | clayg: i like the idea that the diskfile impl knows to create a .durable or not | 18:46 |
| * clayg goes to lookup "lifing a stone" | 18:46 | |
| acoles | clayg: 'opening a can or worms' | 18:46 |
| acoles | s/or/of/ | 18:46 |
| clayg | acoles: so how can I help - i agree it's a good idea - we're just trying to figure out the right way to do it? | 18:47 |
| clayg | acoles: honestly thinking about what's going to make it and what's going to slip for the EC beta - the part I'm most concerned about is - we don't break anything that works today with replication | 18:47 |
| acoles | clayg: what i meant was, i now noticed that we do our hash_cleanup_listdir after writing the .data file, which with EC will do zero cleaning up because the latest durable isn't there yet | 18:48 |
| clayg | if mutli-range GETs and COPYs don't work, if the reconstructor doesn't scale, these are all bad - but if we can iterate on the EC Policy sandboxed off where we won't hurt replicated storage policies - that's a win | 18:48 |
| acoles | clayg: so really the hash_cleanup_listdir should be delayed til durable is written | 18:48 |
| clayg | acoles: hell yeah it should! | 18:48 |
| clayg | acoles: I have another idea on two-phase PUT acctually :\ | 18:49 |
| acoles | clayg: which is more making me feel like an EC-diskfile impl makes sense | 18:49 |
| *** mmcardle has quit IRC | 18:49 | |
| clayg | YES +1 EC-diskfile makes sense | 18:49 |
| clayg | we didn't want to have to do it - but that was before we knew about fragment indexs and durable markers | 18:50 |
| acoles | clayg: ok we're just super-agreeing with each other. and yes, i get the need to land working stuff before worrying too much about 'elegance' | 18:50 |
| clayg | acoles: no i'm in the other camp - i'd rather EC beta barely work and have a super solid framework to iterating on policies than have a crappy mess on master (and an EC beta that mostly sort of works if you're careful) | 18:51 |
| acoles | clayg: ok i misunderstood! | 18:51 |
| * notmyname agrees with what clayg just said | 18:52 | |
| * clayg is shocked! | 18:52 | |
| clayg | thanks for having my back notmyname! | 18:52 |
| clayg | like really if people can benchmark PUT's and GET's and start trying the reconstructor at scale I think we have a good beta | 18:53 |
| acoles | notmyname: clayg: and i didn't disagree, just misunderstood :D | 18:53 |
| notmyname | acoles: oh, I met Jim Meyer this week. told him that it's really great working with you and donagh | 18:53 |
| clayg | I also think we're going to find issues no matter how hard we try - so we may as well get ourselves ready to fix those bugs without a bunch of refactoring and cleanup looming over our heads | 18:53 |
| acoles | notmyname: thanks! | 18:54 |
| clayg | acoles: so if you wanna play with the DiskFileHelpers and see what shakes out that's very helpful | 18:54 |
| acoles | clayg: ok, will do. Now, can i move to a really specific question? | 18:54 |
| clayg | acoles: if you want to try and teach obj.server.ObjectController.get_diskfile how to route to policy specific DiskFiles - that'd be super helpful | 18:54 |
| clayg | acoles: my favorite kind! | 18:55 |
| acoles | ok here goes - agh, all my lights just went out | 18:55 |
| clayg | oh no! | 18:55 |
| *** fbo has joined #openstack-swift | 18:55 | |
| clayg | peluse: oh... i guess you're traveling... anyway i'm going to rebase patch 159637 on acoles rebase of his patch that I stole from him | 18:56 |
| patchbot | clayg: https://review.openstack.org/#/c/159637/ | 18:56 |
| acoles | clayg: ok, back on :) so DiskFileWriter does the data file rename and hash_cleanup_listdir in a separate thread | 18:56 |
| jrichli | acoles: I guess you have stayed late | 18:56 |
| clayg | acoles: yeah it do | 18:56 |
| clayg | jrichli: he likes me | 18:56 |
| acoles | clayg: not really understanding *why* that is, but should the durable file also be created in a separate thread? | 18:57 |
| acoles | clayg: like, same pattern for both | 18:57 |
| acoles | ? | 18:57 |
| acoles | jrichli: clayg : i have to move every 20 mins after 6pm to trigger the light sensors | 18:58 |
| clayg | acoles: i had no idea onus was a werd - sometimes i feel like I'm a total bafoon surrouned by the likes of you luminaries | 18:58 |
| jrichli | we have auto lights going off at 6, but we just push a button to turn them back on for a long time | 18:58 |
| acoles | 'move' as in, roll my chair across the corridor | 18:58 |
| clayg | acoles: doesn't the fsync go down in there? I think the idea is that blocks the hub | 18:58 |
| clayg | os.listdir is blocking to if you're not keeping inodes in ram | 18:59 |
| acoles | clayg: yes, the fsync is done there | 18:59 |
| clayg | acoles: ok - yeah so that's definately why | 18:59 |
| acoles | clayg: but i think we need to fsync the .durable too... | 18:59 |
| clayg | acoles: i think we'd *want* to - yeah | 19:00 |
| acoles | so if i was to either (a) fsync the durable or (b) call hash_cleanup_listdir after writing durable, i should use a separate thread? | 19:01 |
| clayg | acoles: yeah... idk - so doing two hash_cleanup_listdirs seems wasteful | 19:03 |
| clayg | acoles: but we probably want to fsync the .data before we send the continue message | 19:03 |
| clayg | fsyncs (and probably the listdir) should be in a thread to avoid blocking the reactor | 19:03 |
| acoles | clayg: no, i'm thinking an ec-diskfile would defer the cleanup until after the durable is written, so just one hash_cleanup still | 19:04 |
| clayg | acoles: also just FYI - i tried sticking an exception error - maybe an oserror - something in those threads and the object server was not too happy about it - I didn't split off to look at it - but I'm pretty sure it's terrible | 19:04 |
| acoles | clayg: yeah, and we do want to fsync .data before the continue1 | 19:05 |
| clayg | acoles: yeah i'm down with that - i'm just saying when you rename the - yeah you got it | 19:05 |
| acoles | clayg: hmm, i was also wondering what happened to OSErrors down in those threads :/ | 19:05 |
| clayg | acoles: i think the whole world breaks and everything goes to crap - basically what redbo and dfg are always complaning about | 19:06 |
| clayg | we'll probably finally get around to fixing it around the time they finally publish their go implementation :P | 19:07 |
| acoles | lol | 19:08 |
| redbo | awww | 19:08 |
| *** tgohad has joined #openstack-swift | 19:09 | |
| clayg | redbo: i'm just joshin' with ya | 19:12 |
| acoles | clayg: ok i think i need to go and play with the code some more - that was helpful, thanks! | 19:12 |
| *** tsg_ has quit IRC | 19:13 | |
| clayg | acoles: i'm not helping - YOU'RE helping | 19:13 |
| *** silor1 has joined #openstack-swift | 19:14 | |
| acoles | clayg: oh ok, mutual hugs and kisses :P | 19:15 |
| acoles | well, hugs. | 19:15 |
| *** silor has quit IRC | 19:16 | |
| acoles | clayg: notmyname : peluse : fyi i'll be quiet til Monday - taking time off for mother visiting | 19:16 |
| clayg | lol | 19:17 |
| clayg | ok - have a nice weekend | 19:17 |
| *** silor1 has quit IRC | 19:21 | |
| *** acoles is now known as acoles_away | 19:23 | |
| *** bkopilov has quit IRC | 19:27 | |
| *** thumpba_ has quit IRC | 19:30 | |
| notmyname | redbo: dfg: please take a look at https://review.openstack.org/#/c/126923/ | 19:30 |
| *** bkopilov has joined #openstack-swift | 19:32 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Multiple fragment Archive Index Support https://review.openstack.org/159637 | 19:32 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Prefer policy to policy_index https://review.openstack.org/163690 | 19:32 |
| *** nermik has quit IRC | 19:35 | |
| clayg | mattoliverau: i missed some of your comments on patch 163690 because acoles_away had done the rebase - but then it looks like he caught most of them - except for the import thing - which I would avoid because of the rebase pain | 19:36 |
| patchbot | clayg: https://review.openstack.org/#/c/163690/ | 19:36 |
| clayg | mattoliverau: maybe shortly after we cut Kilo we could clean everything up - but w/o a hacking rule it'll just get all mixed up again :P | 19:36 |
| clayg | anyone think we should try H302!? | 19:39 |
| *** openstackgerrit has quit IRC | 19:40 | |
| clayg | i wish abandonded changes didn't still show up in the Needed By dropdown - wtf good is abandonded if it doesn't get rid of them?! | 19:41 |
| *** openstackgerrit has joined #openstack-swift | 19:41 | |
| redbo | I don't really have an opinion on that. I don't think not fsyncing directories has caused us much grief, but from what I've been able to tell the performance drop isn't huge. | 19:44 |
| *** os1 has joined #openstack-swift | 20:01 | |
| os1 | Hi | 20:01 |
| os1 | From swift-recon output, | 20:02 |
| os1 | What do the fields "Failed: ", "no_result", and "reported: " signify? | 20:02 |
| clayg | os1: I'd guess it means timeout vs here's the data? | 20:12 |
| clayg | os1: but if you're looking at it in context you may have a better guess atm | 20:13 |
| clayg | _ahale: you guys have some scripts built up around recon ya - do you use it a lot? we mostly just gather up the raw output from recon dumps into collectd/graphite guages | 20:14 |
| *** mahatic has quit IRC | 20:14 | |
| *** dencaval has quit IRC | 20:15 | |
| os1 | Where's the file where these are defined/explained ? | 20:19 |
| clayg | os1: http://docs.openstack.org/developer/swift/admin_guide.html?highlight=telemetry#cluster-telemetry-and-monitoring mabye? | 20:21 |
| *** nellysmitt has quit IRC | 20:22 | |
| _ahale | oh yeah i can never remember what that means, maybe the no result was for recon-cron stuff thats not run.. we mostly dump stuff to graphite too | 20:24 |
| *** rdaly2 has quit IRC | 20:26 | |
| *** doxavore has joined #openstack-swift | 20:27 | |
| doxavore | i've turned down the object-auditor to do at most 10 files/sec and 1MB/sec, but it still saturates my [new] 7200rpm SATAs (according to atop and iotop) - are there any other recommended settings, perhaps an ubuntu default, that people are known to change to help that along? :-/ | 20:29 |
| doxavore | saturate (v) - iotop and atop report the drive being audited as "100% busy" | 20:29 |
| doxavore | simple dd tests on the drives do over 100MB/s | 20:30 |
| sweeper | hmmm | 20:31 |
| doxavore | sweeper: shhh. i don't look as brilliant when coworkers find i crowdsource all my problems :> | 20:32 |
| clayg | zero_byte_files_per_second maybe? | 20:32 |
| sweeper | don't have a whole lot of those... | 20:34 |
| os1 | Do you know which code file the swift-recon outputs are defined? | 20:35 |
| os1 | The docs are high level about it. | 20:35 |
| clayg | sweeper: well it's an audit function that cleans up for rsync failures - it doesn't do the md5 checksumming - it just checks if the metadata content-length matches the os.stat size | 20:37 |
| clayg | sweeper: doxavore: I just know it runs a seperate rate, it's weird you're getting 100% saturated - if you keep turning the knobs down will it *eventually* not say the drive is 100% busy? | 20:38 |
| clayg | I think maybe there's a mis-reporting thing - like 100% doesn't mean what I'd think it means intuatively | 20:38 |
| clayg | can you vmstat the drives and see if the bytes in/out matches your settings? | 20:38 |
| sweeper | mmm, that's a point | 20:39 |
| clayg | I think *maybe* those settings can make it do less work per unit time - but never acctually doing *nothing* | 20:39 |
| clayg | so it's 100% of the time doing *something* but it could acctually be doing *more* something in that unit time - maybe? i'm splitballin' - it's a strange observation - i've not used those tools specificially. | 20:39 |
| doxavore | atop has typically matched up with vmstat for me historically, but i will certainly take a look at that | 20:40 |
| *** thumpba has joined #openstack-swift | 20:50 | |
| *** thumpba_ has joined #openstack-swift | 20:53 | |
| *** thumpba has quit IRC | 20:56 | |
| os1 | Asked this here earlier, but does anyone know which code file defines/explains what the fields "Failed: ", "no_result", and "reported: " outputted by swift-recon, mean? | 20:58 |
| os1 | Thank you. | 20:58 |
| clayg | swift.cli.recon | 20:59 |
| *** cdelatte has quit IRC | 21:05 | |
| *** vinsh has joined #openstack-swift | 21:09 | |
| *** thumpba_ has quit IRC | 21:15 | |
| *** thumpba has joined #openstack-swift | 21:16 | |
| *** doxavore has quit IRC | 21:23 | |
| os1 | clayg : Could you clarify? I'm not sure where that is pointing to. | 21:29 |
| *** tgohad has quit IRC | 21:29 | |
| os1 | clayg : I see. Thank you. | 21:34 |
| *** cdelatte has joined #openstack-swift | 21:35 | |
| *** delattec has joined #openstack-swift | 21:35 | |
| *** NM has quit IRC | 21:39 | |
| srsakhamuri | notmyname: I am looking for enabling oslo message bus notifications from swift, is using slogging framework a good approach to address this? | 21:45 |
| notmyname | srsakhamuri: :-) I'm working on getting through my emails. can I answer that in a reply. I think I can do that later this afternoon | 21:46 |
| srsakhamuri | ok cool, thanks was wondering whether my mail went to spam :-) | 21:46 |
| *** jkugel has quit IRC | 21:47 | |
| notmyname | srsakhamuri: no. I've been traveling this week. back in the office today with about a 100 things going on at once :-) | 21:47 |
| srsakhamuri | I can understand, take your time. thanks again | 21:48 |
| *** thumpba has quit IRC | 21:50 | |
| *** thumpba has joined #openstack-swift | 21:51 | |
| openstackgerrit | Merged openstack/swift: Bring EC API calls in sync with PyECLib 1.0 https://review.openstack.org/163533 | 21:52 |
| *** annegentle has quit IRC | 21:56 | |
| *** vinsh has quit IRC | 21:59 | |
| *** vinsh has joined #openstack-swift | 22:00 | |
| clayg | what does slogging (bad idea) have to do with oslo message bug notifications (terrible idea)? | 22:00 |
| mattoliverau | Morning | 22:00 |
| clayg | well... "bad" might be too strong of word - some people might see it as more attractive than rolling your own | 22:01 |
| clayg | it'd be better if slogging was better - maybe send notmyname a big 'ol box of round tuits | 22:01 |
| clayg | or just use slogging and keep making patches till it's good enough - it's no so terrible - i used to hack on slogging - i think we could have made it work :P | 22:01 |
| clayg | glange: how far along did you guys get into the message thing before you decided to abandon slogging - I think I was on CBS at the time | 22:02 |
| clayg | at least slogging is a good starting point | 22:03 |
| clayg | any form of proxy log processing and and account data walking is better than olso messages | 22:03 |
| *** vinsh has quit IRC | 22:04 | |
| notmyname | slogging is great. recently dmorita send a bunch of patches and now it works with current swift | 22:06 |
| notmyname | ;-) | 22:06 |
| clayg | well that's huge! | 22:06 |
| clayg | :P | 22:06 |
| notmyname | it ran RAX cloud files billing for years and still runs HP cloud billing. so that's something | 22:07 |
| notmyname | but yeah. could be faster. could be refactored some. could be a lot of things | 22:08 |
| clayg | notmyname: it's nice that it's open source - it'd suck if there wasn't *something* out in the wild | 22:08 |
| clayg | it always just ends up being more efficient to roll your own code because the *big* question is what you're going to do with the data | 22:09 |
| notmyname | if only the author would actually spend any time on it :-) | 22:09 |
| clayg | slogging dumping to csv sorta punts on the issue - which was sort of a design decision IIRC | 22:09 |
| notmyname | ya. which is where simple-billing at RAX came from (the part of slogging extracted to just get the billing metrics and do the calls to the internal billing system) | 22:10 |
| *** tsg has joined #openstack-swift | 22:12 | |
| clayg | oh yeah! I forgot about the SOAP stuff! | 22:13 |
| torgomatic | I step away for 10 minutes and we're talking about SOAP? | 22:16 |
| torgomatic | !? | 22:16 |
| openstack | torgomatic: Error: "?" is not a valid command. | 22:16 |
| torgomatic | that should either be a help command or it should just say "calm the heck down" | 22:16 |
| jrichli | lol | 22:22 |
| *** jrichli has quit IRC | 22:24 | |
| *** logan2 has quit IRC | 22:26 | |
| mattoliverau | lol | 22:28 |
| torgomatic | so, question for everyone: do you know of any application that uses multi-range GET requests to Swift? | 22:40 |
| torgomatic | that is, GET requests that ask for multiple byte ranges, e.g. Range: bytes 1000-5000,178953-199185,... | 22:41 |
| notmyname | torgomatic: is there anything in a proxy log line which would make it a deterministic question? | 22:41 |
| torgomatic | notmyname: not that I know of; single-range GET response and multi-range GET response are both 206 | 22:41 |
| *** thumpba has quit IRC | 22:51 | |
| *** thumpba has joined #openstack-swift | 22:51 | |
| *** annegentle has joined #openstack-swift | 22:56 | |
| *** Nadeem has quit IRC | 22:59 | |
| *** erlon has quit IRC | 23:01 | |
| *** annegentle has quit IRC | 23:02 | |
| openstackgerrit | Ricardo Ferreira proposed openstack/swift: Created iterator for the metadata values, outputs the sysmeta and usermeta separately, each entry in a new line for grep parsing https://review.openstack.org/164019 | 23:25 |
| *** os1 has left #openstack-swift | 23:39 | |
| *** david-lyle is now known as david-lyle_afk | 23:41 | |
| *** kei_yama has joined #openstack-swift | 23:43 | |
| *** thumpba has quit IRC | 23:52 | |
| *** thumpba has joined #openstack-swift | 23:52 | |
| *** reed has quit IRC | 23:55 | |
| *** gyee has joined #openstack-swift | 23:56 | |
| *** annegentle has joined #openstack-swift | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!